PoCoS - Potsdam Coreference Scheme

Summary

Coreference annotation suffers from subjective judgments and from schemes that make incompatible decisions, which makes it hard to compare annotated data across languages. Krasavina and Chiarcos present PoCoS (Potsdam Coreference Scheme). It sets out the general principles that should come before any concrete coreference guidelines. The scheme rests on four requirements: CONSTITUENCY, COMPLETENESS, CONSISTENCY and a preference for MAXIMAL ANALYSIS. From these the authors derive language-neutral rules for choosing markables and antecedents. Following the MATE/GNOME idea, PoCoS separates a general, reusable Core Scheme from more detailed Extended Schemes. It also divides markables into primary ones, which can act as anaphors, and secondary ones, which can only serve as antecedents. A distinctive feature is that ambiguity is marked explicitly and resolved through ranked heuristics. The scheme was instantiated for German, English and Russian newspaper texts. A small agreement study on German data (PCC) gave reasonable kappa scores. The work matters because it makes coreference annotation more portable and comparable across languages, including for parallel corpora.

Key points

Markable rules: a markable covers its head with all of its modifiers (maximum size principle, as opposed to the minimal-string approach of MUC). Function words such as determiners and adpositions are included, so PPs are preferred over NPs.
Antecedent rules: each mention points to the most recent previous mention, forming an ordered chain. Possessive pronouns bound by a controller in the same sentence are an exception. Plural references to several separate antecedents are handled on a separate "groups" layer.
Ambiguity handling: annotators prefer readings that create or extend anaphoric chains (maximize anaphoric chains, primacy of anaphora). Ambiguous cases are marked explicitly, which the authors describe as a novel feature.
Agreement on German PCC data: kappa was 0.69 (union) and 0.82 (intersection). After two annotators independently corrected 13 texts, it rose to 0.76 and 0.89.
Russian adaptation: because Russian lacks articles, secondary markables there are limited to certain quantified expressions. A Russian corpus annotated with PoCoS was still under construction at the time.

Our summary of the paper, in our own words — not its text. The work belongs to its authors; read it at the links above.

Fields: Linguistics & languages

Topics: Coreference resolution

Cited by

15 papers of our library cite it (of the 140673 whose bibliographies we read); 3 of them have a page here, 12 not yet.

Cite

The paper itself, not this page. BibTeX from the library's record of the work.

@inproceedings{krasavina2007pocos,
  title = {{PoCoS - Potsdam Coreference Scheme}},
  author = {Krasavina, Olga and Chiarcos, Christian},
  booktitle = {Proceedings of the Linguistic Annotation Workshop},
  year = {2007},
  url = {https://aclanthology.org/W07-1525/}
}

References

19 references, 6 of them with a card of ours. ↗ leads to the work, «find» searches for one we have no id for yet.

Artstein, R. and М. Poesio. 2005. Kappa 3 =Alpha (or Beta). Technical Report CSM-437, Univ. of Essex. find
Carlson, L., D. Marcu and M. E. Okurowski. 2003. Building a Discourse-Tagged Corpus in the Framework of Rhetorical Structure Theory. Current directions in discourse and dialogue, Kluwer. find our card
Deemter van, K. and R Kibble. 1999. What is coreference, and what should coreference annotation be? Proc. of the ACL Workshop on Coreference. find our card
Ge, M, J. Hale, and E. Charniak. 1998. A statistical approach to anaphora resolution. Proc. of the Sixth Workshop on very Large Corpora. find our card
Holler, A., J.F.Maas and A.Storrer. 2004. Exploiting Coreference Annotations for Text-to-Hypertext Conversion. Proc. of LREC 2004, Lissabon, Portugal. find our card
Hirschman, L. 1997. MUC-7 coreference task definition. Version 3.0. find our card
Karttunen, L. 1976. Discourse referents. Syntax and Semantics. J. McCawley, New York Academic Press. find
Krasavina, O., Ch. Chiarcos and D. Zalmanov. 2007. Aspects of topicality in the use of demonstrative expressions in German, English and Russian. Proc. of DAARC-2007, Lagos, Portugal, 29-30 March. find
Kučová, L. and E. Hajičová (2004). Prague Dependency Treebank: Enrichment of the Underlying Syntactic Annotation by Coreferential Mark-Up. The Prague Bulletin of Mathematical Linguistics 81. find
Leech, G. and J. Svartvik. 2003. A communicative grammar of English. London [u.a.]. find
Leech, G. and A. Wilson. 1996. EAGLES Recommendations for the Morphosyntactic Annotation of Corpora. www.ilc.cnr.it/EAGLES/annotate/annotate.html Mitkov, R. 2002. Anaphora resolution. London [u.a.]. find
Naumann, K. 2006. Manual for the Annotation of in-document Referential Relations. http://www.sfs.uni-tuebingen.de/de\_tuebadz.shtml (July 2006). ↗
Passonneau, R. 1996. Instructions for applying Discourse Reference Annotation for Multiple Applications (DRAMA). Unpublished document. find
Poesio, M. 2004 The MATE/GNOME Proposals for Anaphoric Annotation, Revisited. Proc. of SIGDIAL. find our card
Poesio, M. and R. Artstein, 2005. Annotating (Anaphoric) Ambiguity. Proc. of Corpus Linguistics-05. find
Popescu-Belis, A., L. Rigouste, S. Salmon-Alt, and L-Romary. 2004, Online Evaluation of Coreference Resolution. Proc. of LREC 2004. find
Rocha de, M. 1997. Supporting anaphor resolution with a corpus-based probabilistic model. Proc. of the ACL'97 workshop on Operational factors in practical, robust anaphora resolution. Madrid, Spain. find
Stede, M. 2004. The Potsdam Commentary Corpus. Proc. of ACL-04 Workshop on Discourse Annotation, Barcelona, July. find
Versley, Y. 2006. Disagreement Dissected: Vagueness as a Source of Ambiguity in Nominal (Co-) Reference. Proceedings of the ESSLLI 2006 Workshop on Ambiguity in Anaphora. find

↑ back to top